AI Circular Investment, Like Customer Concentration, is Hard to Avoid

Nvidia's 9.3-percent ownership of Nebius is an example of the AI "circular strategy,” where participants in the AI value chain invest in each other while committing to large-scale purchases of one another's products and services.

Other examples include:

• Microsoft-OpenAI: Microsoft invested over $13 billion (major tranche in 2023). OpenAI became a major Azure customer, with commitments like $250 billion in cloud services.

• Nvidia’s role: Nvidia has invested heavily in OpenAI (up to $100 billion cited), xAI, Mistral, and others, with recipients committing to buy its GPUs. Nvidia also invested in neocloud providers like CoreWeave.

• Amazon/Anthropic/Google: Investments in Anthropic paired with commitments to use AWS, Google Cloud/chips.

• Deals involving Oracle, AMD, BlackRock, etc., including data center acquisitions and massive purchase commitments (Oracle buying Nvidia chips for OpenAI facilities). These form a "tangled web" or " virtuous circle" of financing and buying, depending on one’s perspective. 

The practice is not as unprecedented as it might seem.

Similar strategies can be seen in  prior computing and tech infrastructure investment periods. 

One example is vendor financing. where suppliers extend credit, loans, or equity-like support to customers to enable purchases.

And though the precedent will worry some, who see potential for excess investment, such arrangements were common during the optical fiber investment boom around the turn of the century. 

Equipment vendors helped were active investment partners: 

• Lucent, Nortel, Cisco, Alcatel provided vendor financing (loans, credit, sometimes over 100 percent of purchase value) to buyers building fiber networks

• "Capacity swaps,” where service providers purchased capacity from each other and each partner booked the revenue, also occurred. 

But the computing industry has often seen such deals:

• Enterprise computing suppliers (IBM, later Dell, HP, Cisco) have long used financing, leasing, and channel programs to help customers acquire servers, networking gear, and software

• In the personal computing era, vendor financing and trade-ins helped drive volume

• In semiconductors and enterprise IT, suppliers often finance customers to secure market share, especially during technology transitions.

It’s one sort of risk, to be sure. 

But customer concentration sometimes cannot be avoided. 

In fact, capital-intensive industries tend to produce a rule of three structure, where just a few market leaders exist. In the AI market, such concentration might be unavoidable.

The "Big Three" cloud providers (AWS, Azure, Google) represent more than 60 percent of the global cloud infrastructure market. 

For a major supplier such as Nvidia, that translates to a concentrated customer base. In recent quarters, just four customers (hyperscalers) accounted for 61 percent of revenue. 

The biggest customers have represented 50 percent or more of business in some periods. Just six customers drove 85 percent of revenue in one reported quarter.

Suppliers and investors ideally want broad bases to mitigate risk—if one customer cuts spending, others can offset it. Here, the top buyers are interdependent (via partnerships, investments, and shared ecosystems) and move in similar cycles driven by AI progress and monetization.

Customer concentration is always considered a risk. In the case of AI, it is almost unavoidable. Nor should that be surprising in any capital-intensive business in an early stage of development. 

But even long term, a market led by just a few firms  is almost certain to develop.

Compute Cash Flow: Infrastructure Tends to Lead Other EcosystemRevenue Streams

Some observers rightly note that the “time to revenue” for high-performance computing “as a service” suppliers is crucial. At the moment, for example, some would point to service supplier investments to create the infrastructure.

Some might characterize the cash flow as benefitting chip suppliers at the expense of high-performance computing suppliers, and it is hard to argue with that observation. 

But infrastructure creation first; revenues second is a classic pattern in computing. That pattern has been seen in semiconductors, telecommunications, cloud computing, internet infrastructure, and now AI infrastructure.

Yes, there always is risk, as returns are not guaranteed. 

source: Yahoo Finance 

But it is simply a fact that many computing businesses require supply-leading infrastructure. 

Firms must build capacity before customers can fully exploit it, and that investment often involves high fixed costs, before applications demand and use cases can emerge. 

Principle	Explanation
High fixed costs	Infrastructure requires enormous upfront investment before any customers are served.
Low marginal costs	Once infrastructure exists, serving additional customers becomes relatively inexpensive.
Long construction cycles	Fabs, fiber networks, and data centers often require years to build.
Induced demand	Lower prices and better performance stimulate entirely new applications.
Network effects	Infrastructure becomes more valuable as additional users and complementary services appear.
Option value	Excess capacity allows entrepreneurs to create products that previously were impossible or uneconomic.

So will there be a revenue lag for high-performance computing utilities? Yes. 

Era	Infrastructure investment	Revenue followed later through...	Time lag
Mainframes (1960s)	IBM manufacturing plants, semiconductor production, service organizations	Enterprise computing adoption	Several years
Semiconductor fabs (1970s-present)	Multi-billion-dollar fabrication plants	PC, mobile, cloud, AI chip demand	2–5 years
Personal computers (1980s)	Intel processor fabs, Microsoft software ecosystem, OEM manufacturing	Mass PC adoption	3–5 years
Internet backbone (1990s)	Fiber optic networks, routers, submarine cables	E-commerce, search, streaming	5–10 years
Mobile broadband (2000s)	3G/4G towers, spectrum, fiber backhaul	Smartphone economy, app stores	3–8 years
Hyperscale cloud (2006-present)	Massive global data centers	Cloud software, SaaS, AI services	3–10 years
Content delivery networks	Global edge server deployments	Video streaming and cloud gaming	Several years
AI infrastructure (2023-present)	GPU clusters, AI factories, power generation	AI agents, enterprise AI, robotics, scientific computing	Still developing

But that is a documented pattern:

• Chip fab investment precedes chip sales revenue

• Data center investments preceded cloud computing as a service revenues

• Railroads preceded nationwide commerce

• Electric grids preceded widespread electrification

• Fiber preceded streaming

So building GPU clusters will precede AI-native businesses.

But apparent overinvestment will always be a concern. 

Historically, such overinvestment often also occurs, whether that is long-haul fiber; dynamic random access memory, solar panels or data centers. 

The dangers of overinvestment also are real. But the necessity of infrastructure investment before applications, use cases and revenue can develop is a reality we have often seen. 

For Every Public Policy There are Corresponding Private Interests

I learned a long time ago, as a student of public policy and then as a journalist, that “for every public policy there are corresponding private interests.”

So arguments about whether and how to regulate artificial intelligence in the context of content businesses always will be a combination of abstract public values, impact on culture, fairness, content quality or art and perceived personal economic interest.

Every major technological shift affecting content industries has altered

• who creates value

• who captures income

• whose social status changes. 

Indeed, much of the economic value in content industries rests on scarcity, and AI threatens to create abundance. That might be a favorable outcome for content consumers, but might harm professional content producers. 

Traditional source of scarcity	Effect of AI
Skilled illustration	AI greatly expands supply
Copywriting	Near-zero marginal production cost
Translation	Instant multilingual capability
Stock photography	Synthetic images substitute for many uses
Voice acting	Synthetic voices compete in many applications
Video production	Increasing automation reduces labor inputs
Software documentation	AI drafts much routine material
Marketing content	Mass personalization becomes inexpensive

Whenever scarcity declines, prices usually follow. So the content industry advocates concern about AI "ethics” also are about protecting existing economic rents. 

That isn’t unusual. All professional associations, licensing requirements and unions, whatever their stated purpose (“safety,” often), are also about protecting economic rents.

Public policy concern	Corresponding private interest
Copyright protection	Licensing revenues
Artist consent	Control over monetization
Transparency	Ability to distinguish human work in the market
Watermarking	Preserve premium pricing for human-created work
Fair compensation	Maintain existing wage levels
Quality concerns	Preserve professional gatekeeping
Educational concerns	Preserve demand for traditional instruction
Safety regulation	Increase barriers to entry favoring incumbents
Cultural preservation	Preserve existing creative institutions

We can cite many content industry examples.

Technology	Incumbents defending existing value	Public argument	Private interest
Printing press	Scribes	Accuracy, religious authority	Preserve copying profession
Photography	Portrait painters	Artistic standards	Maintain commissions
Recorded music	Live performers	Artistic integrity	Preserve performance income
Radio	Newspapers	Media concentration	Advertising revenues
Television	Movie theaters	Cultural effects	Box office
Digital photography	Film manufacturers	Image quality	Film sales
MP3 files	Record labels	Copyright	Music distribution revenues
Streaming	Cable operators	Local programming	Subscription economics
Generative AI	Writers, artists, actors, publishers	Copyright, authenticity, quality	Employment, licensing, bargaining power

Also, AI threatens not only earnings but also professional identity, as creative professions provide:

• expertise

• prestige

• cultural influence

• reputation

• gatekeeping authority

• community standing. 

If AI enables non-experts to produce acceptable commercial work, professionals may lose status even before they lose substantial income.

The broader lesson from economic history is that technological debates are rarely contests between "public good" and "private greed." 

Instead, all public policies have corresponding private interests.

AI raises authentic questions about authorship, consent, cultural diversity, and market power. 

At the same time, it redistributes income, bargaining power, and professional status across the content ecosystem. 

That isn’t to deny the legitimacy of the issues raised. But neither does it make sense to deny the private financial interests also at stake.